Logs errors, corrections, and insights to markdown files so you can actually learn from mistakes instead of repeating them. Creates `.learnings/ERRORS.md`, `LEARNINGS.md`, and `FEATURE_REQUESTS.md` files that capture command failures, user corrections, knowledge gaps, and better approaches as they happen. Works well with OpenClaw's workspace system where learnings get promoted to `AGENTS.md` or `TOOLS.md` for broader use. The structured logging format includes priority levels, timestamps, and action items, making it easy for coding agents to process fixes later. Honest take: most AI workflows lose context between sessions, but this creates a persistent memory that actually improves over time.
npx skills add https://github.com/pskoett/self-improving-agent --skill self-improvement